A second region can serve distant readers and survive a regional outage.
It also adds replication delay and a new ownership problem. If both regions accept conflicting writes while disconnected, you need a merge rule that preserves the business invariant. A last-writer-wins policy based on local clocks is rarely enough for money or inventory.
State your recovery point objective, the amount of acknowledged data you may lose, and your recovery time objective, how long restoration may take. Asynchronous replication trades a potential loss window for lower write latency. Failover must fence the old writer and route clients to a consistent authority. You'll separate global reads from regional write ownership and decide which work can remain available during a split.